A genome-wide survey of R gene polymorphisms in Arabidopsis

Abstract

We used polymorphism analysis to study the evolutionary dynamics of 27 disease resistance (R) genes by resequencing the leucine-rich repeat (LRR) region in 96 Arabidopsis thaliana accessions. We compared single nucleotide polymorphisms (SNPs) in these R genes to an empirical distribution of SNP in the same sample based on 876 fragments selected to sample the entire genome. LRR regions are highly polymorphic for protein variants but not for synonymous changes, suggesting that they generate many alleles maintained for short time periods. Recombination is also relatively common and important for generating protein variants. Although none of the genes is nearly as polymorphic as RPP13, a locus previously shown to have strong signatures of balancing selection, seven genes show weaker indications of balancing selection. Five R genes are relatively invariant, indicating young alleles, but all contain segregating protein variants. Polymorphism analysis in neighboring fragments yielded inconclusive evidence for recent selective sweeps at these loci. In addition, few alleles are candidates for rapid increases in frequency expected under directional selection. Haplotype sharing analysis revealed significant underrepresentation of R gene alleles with extended haplotypes compared with 1102 random genomic fragments. Lack of convincing evidence for directional selection or selective sweeps argues against an arms race driving R gene evolution. Instead, the data support transient or frequency-dependent selection maintaining protein variants at a locus for variable time periods.

Figure 1.

Neighbor-Joining Trees Based on Jukes-Cantor Corrected Silent Sites (Nei and Gojobori, 1986) with 100 Bootstraps, Sorted Based on the Level of Allelic Divergence. At1g12290 (A), At1g17600 (B), At1g27170 (C), At4g33300 (D), At1g63730 (E), At1g64070 (F), At5g04720 (G), At1g53350 (H), At1g63740 (I), At1g65850 (J), At5g11250 (K), At5g17680 (L), At1g59620 (M), At1g33560 (N), At1g56540 (O), At4g26090 (P), At5g44870 (Q), At5g38850 (R), At4g14370 (S), At5g63020 (T), At3g50950 (U), At5g58120 (V), At1g63750 (W), At1g59780 (X), At4g14610 (Y), At2g16870 (Z), and At3g46530 (AA). Bar beneath each tree reflects 0.001 substitutions per site. Clades highlighted in orange contain at least one nonfunctional allele due to mutations resulting in a frameshift or premature stop codon. Clades highlighted in blue and green contain at least one allele with one or more amino acid insertion or deletion, respectively.

Figure 2.

The Allele Frequency Distribution for Synonymous and Nonsynonymous SNPs Observed for the Set of 27 R Genes and the Set of 876 Random Genomic Fragments.

Figure 3.

PCA for 27 R Genes Based on the Correlation Matrix of Eight Summary Statistics. Summary statistics are K_Smax, π, S, Tajima's D, R_h, F_st, number of protein variants for the full sequence, and number of protein variants for the xxLxLxx motif.

Figure 4.

Difference in Silent Nucleotide Diversity between Five R Genes and Their Flanking Fragments Located at Increasing Physical Distances. Fragments that are located to the left and right side of each R gene are indicated with squares and triangles, respectively. Upper and lower 5% tails and median and average values are indicated by polynomial trend lines calculated for 50-kb sliding windows with a 1-kb increment for silent nucleotide differences between the set of 876 random genomic fragments and their nearest neighbors.

Figure 5.

Haplotype Sharing for 12,403 Alleles in a Set of 1102 Random Fragments and 556 Alleles from the Set of 27 R Genes. Haplotype sharing for the random fragments and R alleles is depicted by gray and black circles, respectively. The 95th percentile of the distribution is represented by a black line. Alleles from the six R genes that contain an allele in the top 5% of the distribution are color coded.

Figure 6.

Correlation between Minimum Number of Recombination Events per Base Pair and Nucleotide Diversity. Correlation between minimum number of recombination events per base pair (R_h) and nucleotide diversity (π) is significant for both the set of 27 R genes (black trend line; R² = 0.89; Kendall's τ = 0.56) and the reference set of 876 random fragments (gray trend line; R² = 0.1458; Kendall's τ = 0.38).